LLM token savings benchmark.
Compares whole-tree vs xref-guided context selection by asking Claude the same questions under both strategies and measuring token usage and answer correctness.
Requires clj-http and cheshire on the classpath (available via :dev profile).
LLM token savings benchmark. Compares whole-tree vs xref-guided context selection by asking Claude the same questions under both strategies and measuring token usage and answer correctness. Requires clj-http and cheshire on the classpath (available via :dev profile).
(print-results results)Print a comparison table.
Print a comparison table.
(run-benchmark {:keys [paths model questions]
:or {paths ["src"] questions test-questions}})Run the full benchmark. Returns vector of result maps. Options: :paths - source paths to analyze (default ["src"]) :model - Claude model (default haiku) :questions - override test questions
Run the full benchmark. Returns vector of result maps. Options: :paths - source paths to analyze (default ["src"]) :model - Claude model (default haiku) :questions - override test questions
(whole-tree-context paths)Build context from all source files under paths.
Build context from all source files under paths.
(xref-guided-context db sym paths)Build context from only the files in sym's xref neighborhood.
Build context from only the files in sym's xref neighborhood.
cljdoc builds & hosts documentation for Clojure/Script libraries
| Ctrl+k | Jump to recent docs |
| ← | Move to previous article |
| → | Move to next article |
| Ctrl+/ | Jump to the search field |